Sheng-Jie Luo
Yonghao Yue
Chun-Kai Huang
Yu-Huan Chung
Sei Imai
Tomoyuki Nishita
Bing-Yu Chen
National Taiwan University
Columbia University
National Taiwan University
National Taiwan University
The University of Tokyo
UEI Research / Hiroshima Shudo University
National Taiwan University / UEI Research
Legolization
Optimizing LEGO Designs
LEGO Construction Activity
Courtesy of Robin Sather - Brickville DesignWorks - LEGO Certified Professional Builder
LEGO Sculpture
Courtesy of the LEGO club of The University of Tokyo
LEGO Sculpture
Courtesy of Robin Sather - Brickville DesignWorks - LEGO Certified Professional Builder
LEGO Construction Problem
© SEGA
Legolization
requirements: shape
Legolization
requirements: color
Legolization
requirements: stability
Applications of LEGO Construction
Entertainment/Professional Sculpture
Courtesy of Robin Sather - Brickville DesignWorks - LEGO Certified Professional Builder
Applications of LEGO Construction
Rapid prototyping
[Mueller et al. 2014]
Applications of LEGO Construction
Architecture designs
Previous Approaches
[van Zijl and Smal 2008]
[Gower et al. 1998] [Petrovic 2001]
[Testuz et al. 2013]
Previous Approaches
Articulation pointConnectivity Alignment
Heuristic-based metrics:
...
Problem of Heuristic-based Metrics
Metric of [Gower et al. 1998]
<1.20 x 104 2.10 x 104
Problem of Heuristic-based Metrics
Metric of [Gower et al. 1998]
Metric of [van Zijl and Smal 2008]
Metric of [Petrovic 2001]
Metric of [Testuz et al. 2013]
< 7.30 x 106 2.74 x 107
1.23 x 107 3.13 x 107 1 19
1.20 x 104 2.10 x 104 <
< <
Problem Definition
Problem Definition
Surface voxel
Inner voxel
Problem Definition
Goal: maximal layout
Problem Definition
Goal: physically realizable
Brick Family
1x1
1x2
1x3
1x4
1x6
1x8
2x2
2x3
2x4
2x6
2x8
Fundamental Observation
Per layer:
102 maximal layouts
Total:
1023 maximal layouts
Fundamental Observation
Deterministically enumerate all candidates
and check them one by one.
Randomly generate a layout and hope it will
be satisfactory.
Core Idea: Force-based Stability Analysis
Structure A Structure B
Stability metric (A) < Stability metric (B)
0 < Stability metric (B)
Core Idea: Stability Aware Refinement
Stability
Iteration
Stability
threshold
Bad structures
Good structures
. . .
Layout Initialization
Metric: #components
#components = 2
Stability-aware Refinement
Structure analysis
Layout reconfiguration
Metric
Portion to
refine
Seed brick (structural-critical)
k-ring neighbors (k=1) k-ring neighbors (k=2)
Layout Reconfiguration
Layout
...
Layout Reconfiguration
The value of k:
(1) Want enough chance to visit possible layouts
(2) Close neighborhood are usually sufficient
Current fail count
Constant
Layout Reconfiguration
Region to refine
Brick Operations
(1) Split operation:
Brick Operations
(2) Repeated remerge operation:
Stability-aware Refinement
Structure analysis
Layout reconfiguration
Metric
Portion to
refine
Force-based Stability Analysis
Gravity
When is a layout stable?
At each brick:
1. Forces are balanced
2. Forces are within ranges
Gravity
Snapping between Bricks
Forces Working on Bricks
Forces Working on Bricks
Forces Working on Bricks
Gravity Forces
mg
mg mg
Friction Forces
Friction Forces
Friction Forces
Support Forces
Support Forces
Normal Forces
Normal Forces
Normal Forces
Internal Forces between Bricks
mg
mg mg
Force Balance
Translational equilibrium for brick :
mg
Force Balance
Rotational equilibrium for brick :
mg
One-sided-ness of Forces
Non-negativity condition:
Maximum Friction Load
Fi <T Þ Bricks remain snapped
Hypothetically,
We could try to solve for
This does not give an answer for unstable cases
T > Fi Î Ff
Capacities
For a friction force , define its capacity as
Capacities
The capacity for a sculpture
Stability Analysis
If forces can be redistributed to make ,
the LEGO sculpture is stable
Stable Unstable
++ -
Stability Analysis as an Optimization
Stability Analysis as an Optimization
This problem is solvable for both stable and unstable cases
Maximum Capacity as Stability Metric
Unstable
Threshold for Stability
Stable Unstable
Weakest Portion
Unstable
Results: Verification of Stability Analysis
Results: Verification of Stability Analysis
Results: Verification of Stability Analysis
Results: Performance
Model #Bricks
Layout Init.
Time
Layout Opt.
Time
Assembly
Time
1642 1.543s 17.89s 8h
1706 0.709s 2054s 9h
8277 67.57s 1259s 4d
Results: Performance
Model #Bricks
Layout Init.
Time
Layout Opt.
Time
Assembly
Time
2322 2.347s 16.98s 10h
17755 225.3s 266.1s -
9543 63.09s 108.4s -
Results: Accounting for Brick Number Limit
Results: Accounting for External Weights
Input external force
500g
Without external force
With external force
Results: Building Various LEGO Sculptures
Input voxelized
representation Real assembled sculpture
Results: Building Various LEGO Sculptures
Input voxelized
representation
Assembled
sculpture
Refinement
process
#iterations
-50
-40
-30
-20
-10
0
10
20
0 1 2 3
Results: Building Various LEGO Sculptures
Input voxelized
representation
Assembled
sculpture
Refinement
process
#iterations
-4
-3
-2
-1
0
1
0 20 40
Results: Building Various LEGO Sculptures
Layer 1
Results: Building Various LEGO Sculptures
Layer 2
Results: Building Various LEGO Sculptures
Layer 2
Results: Building Various LEGO Sculptures
…
Results: Building Various LEGO Sculptures
Layer 49
Results: Building Various LEGO Sculptures
Layer 50
Results: Building Various LEGO Sculptures
Results: Building Various LEGO Sculptures
Limitations
Intermediate assembly can still be unstable
-11.49414.323
Limitations
-35
-30
-25
-20
-15
-10
-5
0
5
10
0 50 100 150 200 250 300 350
#iterations
There is no stable layout
Conclusions
Stability analysis for LEGO sculptures
Force-based stability metric
Weakest portion
Conclusions
Automatically generates LEGO brick layouts
Reducing the number of bricks
Satisfies color constraints
Being able to physical stable
Acknowledgements
Christopher Batty, Gabriel Cirio, Anne Fleming,
Eitan Grinspun, Wei-Ting Lin, Han-Wei Liao
Student LEGO Builders from National Taiwan University:
Chin-Yu Chien, Chi-Hao Hsieh, Tsung-Hung Wu, Kang Jao,
Jia-Yu Tsai, Che-Chun Hsu, Fan Wang, Wei-Tse Lee, Yung-Ta Lin,
Li-Ming Yang, Long-Fei Lin, Xiao-Feng Jian, Kai-Han Chang,
Ming-Shiuan Chen
LEGO Certified Professional Builder - Robin Sather
The LEGO Club of The University of Tokyo
Intel, JSPS, MOST
Legolization: Optimizing LEGO Designs
THANK YOU

Legolization: Optimizing LEGO Designs (SIGGRAPH Asia 2015)